Transcriptome sequencing (RNA-Seq) technology has been developed in recent years using a new generation of high-throughput sequencing technology (Marguerat and Bähler 2009). Wilhelm et al. (2008) and Nagalakshmi et al. (2008) used RNA-Seq technology to study the fission yeast and Saccharomyces cerevisiae transcriptome, thus symbolizing the establishment of RNA-Seq technology. RNA-Seq yields abundant information about RNA transcripts with extremely high detection accuracies and enables the discovery of transcripts at lower abundance (Shen et al. 2011, Chen et al. 2018). The most unique application of RNA-Seq is the analysis of differentially expressed genes (DEGs) in different sample transcriptomes (Wall et al. 2009, Gu et al. 2019). The application of transcriptomics is very extensive (Qi et al. 2011, Liu et al. 2019) and includes structural studies of transcripts (border identification of introns, initiation codon identification, alternative splicing studies, and UTR identification), studies on the structural variations of transcripts (identification of fusion genes and study of the polymorphism of coding sequences), gene expression studies (Wang at el. 2010), functional studies of noncoding regions (e.g., studies of noncoding RNAs, microRNAs, and small interfering RNAs) (Clamp et al. 2007, Ponting et al. 2009), the discovery of new low-abundance recordings, and more.

Cao et al. (2013) used RNA-Seq to identify the DEGs of Chironomus kiiensis Tokunaga. In addition, pathway enrichment analysis also showed that these DEGs were annotated to the metabolic pathway. Pan et al. (2015) compared the transcriptomes of thiamethoxam-resistant strains and another susceptible strain of aphids. The results demonstrated that acetylcholine receptor gene changes, upregulated ribosomal protein, ecdysone uridine diphosphate glucose transferase, cytochrome C oxidase, esterase, and peroxidase were the main mechanisms of aphid resistance to thiamethoxam. Following the completion of the whole-genome sequencing of Drosophila melanogaster (Meigen) in 2000, many insect transcriptome sequences have been reported. With the emergence and development of high-throughput sequencing technology, this gene information is an important sequence resource in insect toxicology research.

However, there have been no studies on the transcriptome of Micromelalopha troglodyta (Graeser). Therefore, in this study, the deep transcriptome of M. troglodyta was sequenced by using Illumina HiSeq™ 2000 sequencing technology. Then, the transcriptome was globally analyzed by assembly, annotation, and bioinformatics analysis, including classification, protein function prediction and classification, and metabolic pathway analysis. This study used data from RNA-Seq to obtain gene transcriptional expression levels. A comprehensive analysis of the DEGs was performed using transcriptome data. Furthermore, to understand the biological activity and biological processes of detoxification-related genes, the differentially expressed detoxification-related genes were further identified. This work establishes a foundation for the study of the M. troglodyta genome, which has great theoretical significance for studying genes related to pesticide targets and metabolic processes in M. troglodyta. At the same time, the study results also provide a theoretical basis for the study of the relationship between herbivores and host plants and the extensive interaction between them.

Materials and Methods

Insect rearing and sample preparation. The M. troglodyta population was collected from poplar (Populus × euramericana ‘Nanlin 895′) in Nanjing, Jiangsu Province, China. The larvae were reared in the incubator at 26 ± 1°C, relative humidity of 70–80%, and photoperiod of 16:8 (light:dark). The newly collected poplar leaves were used as food for larvae. Healthy third-instar larvae of a similar weight and appearance were used for the tannin treatments (TTs).

First, the tannin was dissolved with a small amount of ethanol and then was diluted with distilled water to the concentration of 0.1 mg/mL. Newly collected poplar leaves were soaked in the tannin solution for 10 s. After the leaves dried, two treated leaves were placed into a triangular flask with five third-instar larvae. This procedure was repeated with 10 triangular flasks. The control (CK) group consisted of larvae feeding on leaves soaked in distilled water. After feeding on the treated leaves for 96 h, the larvae were collected in groups and dissected. The larvae of M. troglodyte were dissected on ice. After the peritrophic membrane containing midgut contents was removed, midguts were washed in 1.15% ice-cold KCl and collected. All experiments were independently conducted three times.

RNA isolation and sequencing. We used the RNprep Tissue Kit (TIANGEN) to separate the total RNA according to the manufacturer's guidelines. The concentration of RNA was measured with a spectrophotometer, and the integrity of RNA was detected by the agarose gel dissolution method with a concentration of 1%. The total RNA was purified by using poly-T oligo magnetic beads to obtain mRNA, and fragmentation buffer was to interrupt mRNA. First, we used random hexamers to synthesize the first cDNA strand, and then the second cDNA was synthesized with dNTPs, DNA polymer I, and RNaseH. Solexa sequencing using a Illumina HiSeq 2000 instrument was performed at the Shenzhen Huada Gene Research Institute.

Sequence assembly. Trinity were used to perform the de novo assembly and eliminate PCR duplication. Then, when the transcripts were aggregated into unigenes, Tgicl was usually used. Trinity is composed of three single-handed software modules, namely, Inchworm, Chrysalis, and Butterfly, which are successively used to process a large number of reads. Trinity divides the sequence data into a number of separate de Bruijn maps, with each map representing some point of view of the transcription complexity of a given gene or site. Each map is separately assessed to extract the full-length splicing subtype and separate the transcripts from homologous genes.

Multiple samples of the same species were used for sequencing. Sequence clustering software was used to splice and remove the single gene in each sample set so as to obtain the required nonredundant single gene. In the final step, we used Blast to compare unigenes with the five databases (NCBI nonredundant nucleic acid database [NT], NCBI nonredundant protein sequences [NR], Cluster of Orthologous Groups of proteins [COG], Kyoto Encyclopedia of Genes and Genomes [KEGG], and Swissprot) to obtain the result of annotation; Blast2GO with NR annotation was used to obtain the Gene Ontology (GO) annotation (Altschul et al. 1990, Conesa et al. 2005). We, thus, established five samples of M. troglodyta. The unigenes assembled with these six samples were further sequenced and deduplicated using sequence-based clustering software to obtain the longest possible nonredundant unigenes.

Sequence annotation. NR and Swissprot are protein databases, and COG is also a protein database based on complete genomes of bacteria, algae, and eukaryotes that provides a direct homologous classification based on system evolution. KEGG is a database containing gene function and cell metabolism and signaling pathway information. Unigenes were matched with four major protein databases (NR, Swissprot, KEGG, and COG) through BLASTx (evalue, <0.00001). Then, the unigenes were compared to the nucleic acid database NT (evalue, <0.00001). When the comparison between different databases was inconsistent, according to the priority of five databases (NR, Swissprot, KEGG, and COG), the sequence direction of a Unigene was determined. Gene function and metabolic pathway classifications were analyzed by the sequence similarity method, and the unigenes were compared to the GO gene function classification database and KEGG metabolic pathway database to obtain the functional classification and the prediction of metabolic pathways.

Digital gene expression library preparation and analysis. The number of reads for RNA sequencing of different samples differed. If the number of matched reads is considered at the gene expression level only, there is confusion when comparing the expression levels of certain genes among different samples. To resolve this problem, the reads were standardized by matching the genes (Mortazavi et al. 2008) using the reads per kb per million reads (RPKM) value to represent the transcriptional expression level of the gene. The RPKM method successfully eliminates the effect of different gene lengths and sequence differences on the calculation of gene expression levels. Thus, the RPKM can be directly used to compare the differences in gene expression levels between the TTs and the CK. In order to identify the DEGs between TT and CK, the threshold of P value was ascertained by the false discovery rate (FDR) method. Generally, when the threshold FDR was ≥0.8 and | log2ratio | was ≥1, we think that there are significant differences in gene expression. Then, GO enrichment analysis and KEGG pathway enrichment analysis were used to further annotate genes expressed across different stress levels.

Real-time fluorescent quantitative PCR and data analysis. Real-time fluorescent quantitative PCR (qPCR) was conducted with a SYBR® Premix Ex Taq™ II (TliRNaseH Plus) (Takara, Japan) kit in an ABI 7500 instrument (Applied Biological System). Software primer premier 5.0 was used to design gene-specific primers based on gene sequence template (Table 1). The amplification of cDNA by qPCR was performed in a 20-µL mixture that contained approximately 1µL of cDNA, 10 µL of SYBR Premix Ex Taq, 0.4 µL of Rox reference dye, 0.4 µL of both sense primer (10 µM) and antisense primer (10 µM), and 7.8 µL of double-distilled water. Actin was used as an internal standard (0.4 µL for each). The following qPCR procedure was used: 95°C for 30 s; 40 cycles of 95°C for 5 s and 60°C for 34 s; and 95°C for 15 s, 60°C for 1 min, and 95°C for 15 s for plate reading. At the end of each operation, a solution chain curve was produced for each sample to evaluate the purity of the amplified product. All experiments were independently conducted three times. The relative expression level of M. troglodyta mRNA was calculated by using the 2^–ΔΔCT method (Giulietti et al. 2001).

Table 1 Primers used in real-time RT-PCR.

View Fullscreen

Results

M. troglodyta transcriptome assembly. The detoxification of M. troglodyta larvae under tannin stress was studied by RNA-Seq. Reads were assembled using Trinity. After data filtering, 51,797,038–54,991,822 reads for the TT library and 51,674,478–52,307,172 reads for the CK library were acquired. Trinity software was used to further assemble these clean reads into 52,246–55,203 contigs with a mean length of 363–421 bp in the TT, and 46,918–64,401 contigs with an average length of 349–400 bp in the CK were obtained. In our study, 33,008–36,353 unigenes were obtained with a mean length of 574–674 bp for the TT and 29,948–42,194 unigenes with a mean length of 526–638 bp for the CK. There were 46,518 unigenes with a mean length of about 906 bp. They were assembled from clean reads of N₅₀ with a length of 1,568 bp (Tables 2 and 3). After assembly, 23,792 unigenes were annotated in 5 public databases (Table 4). All 21,236 unigenes were grouped with the NR database, 15,655 unigenes were matched with the Swissprot database, 14,032 unigenes were matched with the KEGG database, 7,678 unigenes were matched with the COG database, and 9,893 unigenes were matched with the GO database (Table 4).

Table 2 Statistical output of transcriptome sequencing in M. troglodyta.

View Fullscreen

Table 3 Assembly quality statistics for transcriptome sequencing in M. troglodyta.

View Fullscreen

Table 4 Annotation result statistics for transcriptome sequencing in M. troglodyta.

View Fullscreen

Functional annotation of the M. troglodyta transcriptome. The distribution of the evalues of the identified M. troglodyta unigenes revealed that 30.4% of the unigenes shared the greatest homology with an evalue cut-off of <1e–100 (Fig. 1a). In addition, the semblance distribution of the distinguished unigenes indicated that the semblance of more than 63.0 % of unigenes to their closest homologous genes was greater than 60.0 % (Fig. 1b). Certainly, the greatest number of unigene matches were for the insect genome, and Danaus plexippus (L.) (62.7 %), Bombyx mori (L.) (9.3 %), Papilio xuthus (L.) (3.5 %), Tribolium castaneum (Herbst) (2.9 %), and Helicoverpa armigera (Hübner) (1.5 %) accounted for the top 5 unigenes based on the NR annotations. The rest (20.1 %) of the sequences showed good homology with those of other insects (Fig. 1c).

View Figure

• Download as Powerpoint
• Download Figure

The GO database was commonly used for gene functional annotation (Ashburner et al. 2000). Blast2GO software was used for gene annotation and matched the transcriptome of M. troglodyta to 3 major functional processes, including 59 GO terms. In other words, according to the GO gene functional classification system, 9,893 unigenes were divided into the 3 main functional ontologies—biological process, molecular function, and cellular component (CC) (Fig. 2). In view of the GO analysis, approximately 54.42% of genes were in the biological processes category, and the rest of the genes were in cellular processes (29.06 %) and molecular processes (16.51 %). In biological processes, the main subcategories were cellular process (6,126) and metabolic process (5,005) and next was the single-organism process (4,619). For the cellular component category, cell parts (4,609), cells (4,610), and organelles (3,338) were the most frequently represented. In terms of molecular function, binding (4,954) and catalytic activity (4,928) were highest. Nevertheless, in these three main categories, few genes were allocated to virion, protein tag, and receptor regulator activity.

View Figure

• Download as Powerpoint
• Download Figure

In total, 7,678 sequences were subjected to COG classifications; they were divided into 25 COG groups using WebMGA, with an evalue cut-off of 1e^–5. In the 25 COG classifications, the greatest group was general function prediction (2,899), followed by replication, recombination and repair (1,314) and translation, and ribosomal structure and biogenesis (1,296). This may be related to the fact that there is still currently little data on M. troglodyta in the COG database. In all, 1,089 unigenes with unknown function were obtained by sequencing and are presumed to be new genes specific to M. troglodyta (Fig. 3).

View Figure

• Download as Powerpoint
• Download Figure

DEGs of M. troglodyta in response to tannin stress. To explore the detoxification mechanism to tannin stress in M. troglodyta, detoxification response genes that were up- or downregulated in larvae under tannin stress were identified by using Illumina HiSeq 2000 sequencing. For the purpose of maximizing the accuracy of the measurement of expression levels, merged data from three replicates, including RPKM values, were computed and the results between the replicates for the TT and CK groups were compared (Fig. 4). When FDR was ≥0.8 and |log2Ratio| was ≥1, the difference between TT and CK was considered significant. Among the 23,792 unigenes, a total of 1,627 DEGs were ascertained (Fig. 5). We found that 885 of these genes were upregulated and the other 742 were downregulated.

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

Real-time fluorescent quantitative PCR analysis. In order to further estimate the DEGs identified from the transcriptome library, some DEGs were selected and quantified by qPCR under tannin stress. The results illustrated that the expression level of the selected unigenes was the same as that obtained in M. troglodyta transcriptome data (Fig. 6). The expression of all 18 genes was consistent with the RNA sequence data. It has been reported that actin, which is stably expressed in insects, is an appropriate reference gene for data standardization. In general, the results of qPCR were consistent with the transcriptome data, and the changes detected in mRNA sequencing were confirmed to be true.

View Figure

• Download as Powerpoint
• Download Figure

Differentially expressed detoxification-related genes. Detoxification-related DEGs were determined from the transcriptome library. In our study, the results revealed that tannin stress induced the expression of several different genes involved in detoxification, such as glutathione S-transferase (GSTs), cytochrome P450s (CYP), and uridine diphosphate-glycosyl transferases (UGTs). UGTs are one of the most essential enzymes in phase II reactions, and CYPs are one of the main enzymes in phase I reactions. They also play very critical roles in the decomposition of endobiotics and xenobiotics (Feyereisen 2012, Sun et al. 2019). Our results indicated that TT resulted in the downregulation of 4 CYPs and the upregulation of 26 CYPs in the larvae of M. troglodyta. One UGT was downregulated by 3.83-fold in treated larvae, while three UGTs were upregulated by 3.14- to 4.5-fold under tannin stress. GSTs played a significant role in the insect detoxification process. Interestingly, DEG analysis indicated a series of GST genes that demonstrated different levels of induction or inhibition according to tannin stress. For example, the expression of five GSTs was downregulated in the TT library, whereas five of these genes were upregulated in the TT library (Table 5).

Table 5 Detoxification-related genes associated with tannin stress in M. troglodyta.

View Fullscreen

Table 5 Continued.

View Fullscreen

Table 5 Continued.

View Fullscreen

Discussion

Tannin, also known as gallic acid or tannic acid, widely exists in the roots, stems, leaves, fruits, and bark of plants. Tannin is a polyphenolic compound with a relative molecular weight of 500–3,000 u. According to the solubility of tannin, it can be divided into soluble tannin and insoluble tannin. Soluble tannin is considered the main substance causing astringency, and it combines with human oral mucosal proteins to produce strong astringency. Plant tannins are a type of compound that has been studied previously and frequently in natural products. Many studies have shown that phenolic hydroxyl groups in tannins can interact with enzymes in pathogens and, thus, produce toxicity to the gastrointestinal microorganisms of animals (Goel et al, 2005). In our study, we recognized many kinds of DEGs and signaling pathways involved in the M. troglodyta response to TT.

The theoretical basis of RNA-Seq is as follows: all RNA in a specific cell or tissue is isolated, a cDNA library is constructed, and the cDNA library sequences are randomly fragmented into small fragments. Alternatively, RNA fragmentation can be followed by reverse transcription. Sequencing is performed using a new generation of high-throughput sequencing. The reads are compared to the database (reference genome) or de novo assembled (no reference genome). Finally, a genome-wide transcriptome is formed, and the gene functional annotation, expression annotation, and participating metabolic pathways are analyzed (Morozova et al. 2009). Qin et al. (2011) used the HiSeq 2000 platform for deeply sequencing normal and regenerated tissue in Dugesia japonica Ichikawa & Kawakatsu and then established digital gene expression profiles and transcriptional maps, which provided a broad and deep molecular biology background for the development of this model organism, especially the exploration of genes involved in D. japonica regeneration (Qin et al. 2011). At present, the sequence data of M. troglodyta have not been reported domestically or internationally, but obtaining more information on the M. troglodyta sequence presents a better way to research gene functions in this species. This study used Illumina HiSeq 2000 sequencing technology and annotated the reference transcriptional database of M. troglodyta. Then, 28,365,876,720 bp of data were obtained, marking the first time that the RNA-Seq technology has been used to study and obtain the complete transcriptional information of M. troglodyta. The results of this experiment provide extensive sequence resources for M. troglodyta. Thus, RNA-Seq was used to identify DEGs. In addition, RNA-Seq was used to lay the foundation for further in-depth studies and to systematically and comprehensively define mechanisms of action or resistance.

GST/glutathione metabolism associated with tannin stress. During evolution, plants produce secondary metabolites to protect themselves from phytophagous insects or interfere with their growth and development. Tannins are a large group of polyphenolic compounds and are a natural protective substance for plant survival. To adapt to their ecological environment, insects decompose toxic substances into nontoxic substances mainly by detoxifying enzymes (GSTs, CYP, and esterase), or the toxic substances are used by insects or excreted from their bodies.

Wang et al. (2004) used microarray and genomic techniques to study the Malpighian tubule of D. melanogaster. The researchers found that, in addition to osmotic regulation, the Malpighian tubule also functions in transferring tissue solutes (Wang et al. 2004). Therefore, the Malpighian tubule can excrete a wide range of tissue lysates and xenobiotic biological metabolites. The Malpighian tubule enhances its excretion mainly by expressing CYP enzymes, GSTs, and alcohol dehydrogenase in large amounts (Dow and Davies 2006). These three enzymes play a significant role in the metabolism and detoxification of endogenous lysates and xenobiotic organisms.

GSTs comprise an important metabolic enzyme system in organisms that participates in the primary and secondary metabolism of exogenous substances. GSTs catalyze the nucleophilic reaction of endogenous glutathione with a substrate by the conjugation of glutathione. GSTs mainly transfer a group of electrophilic substrates to the sulfur atom of endogenous reduced glutathione, which makes electrophilic substances hydrophilic and easy to excrete and detoxify (Enayati et al. 2005). The GST gene plays a significant role in the insect detoxification process. It also plays a major role in protecting insects from the reactive chemicals formed by the decomposition of endogenous compounds and the biotransformation of foreign compounds. (Maher 2005, Rinaldi et al. 2002). Furthermore, the GST gene plays an important role in the storage and transport of reduced sulfur, the synthesis of proteins and nucleic acids, the regulation of enzyme activity, the maintenance of the antioxidant properties of tissues, and the regulation of redox-sensitive signal transduction (Yan et al. 2014).

GSTs are widely distributed within living organisms. In mammals, GSTs are mainly distributed in liver microsomal cells, and GSTs are also active in serum. In insects, high levels of GSTs are found in the fat body, digestive tract, and Malpighian tubule. Studies have shown that GSTs are one of the most important enzymes for insects to metabolize insecticides, and they play an important role in resistance to some pesticides, especially organophosphate and carbamate pesticides. In addition, insect GSTs play an important role in insect resistance to plant secondary substances and other exogenous toxic substances (Chen and Gao 2005). li et al. (2010) found that there were 3 specific upregulated expression patterns in 11 GST genes, of which 2 were upregulated in the genomic sigma family, and the expression folds (2.83- and 4.30-fold) were higher than those in the delta family (2.06) when alachlor was used to treat mosquitoes. These results suggest that sigma family GSTs may play a major role in the metabolism of alachlor, which means that the degradation metabolism of alachlor in the mosquitoes mainly uses GSTs of the sigma family (Li et al. 2010). Chen et al. (2005) found that the activity of GSTs in Plutella xylostella (L.) treated with a low concentration of total alkaloids of Tripterygium wilfordii Hook. f. (LC₁₀) was significantly increased in 1st and 2nd instar larvae, reaching 2.83 times that of the CK. These results indicate that T. wilfordii alkaloids can significantly reduce the enzyme activity of GSTs at low concentrations (Chen et al. 2005). Gao et al. (1999) found that the activity of GSTs increased 4–8 times after feeding H. armigera with artificial diets containing 0.01% rutin, 2-tridecanone, and quercetin. After induction of the 2nd generation of H. armigera with 0.01% quercetin, it was found that the GST activity of the induced population was increased by nearly 15 times, and the activity of carboxylesterase was increased by 2 to 3 times. These results indicate that quercetin can induce overexpression of GSTs in H. armigera, suggesting that GSTs are one of the most important detoxification enzymes in H. armigera (Gao et al. 1999). These studies demonstrate that individual or multiple GST genes show significant changes under stress from insecticides or plant secondary metabolites.

We used transcriptomes to analyze the upregulation or downregulation of all GST genes in M. troglodyta. We selected 11 DEGs for GSTs from the transcriptome. For example, five GST genes were downregulated in the TT library, and five of these genes were upregulated in the TT library. We also found that 11 DEGs were involved in the glutathione metabolism pathway. GST is an important enzyme in glutathione metabolism. In addition to functioning in intracellular toxic substance binding, transport efflux, and glutathione reductase, GST also protects cells from oxidative damage. This implies that M. troglodyta may reply to tannin stress by increasing the expression of GSTs, which may illustrate the detected activation of GSTs under tannin stress.

Drug metabolism CYP associated with tannin stress. The CYP protein binds to CO catalyzed by Fe²⁺ and has a characteristic absorption peak at 450 nm, so it is named the CYP enzyme system. CYP enzymes are 46-to 60-kDa proteins with similar structures and different properties. CYP is an important oxidase system located in the smooth endoplasmic reticulum. This enzyme can synthesize and degrade insect pheromones and hormones and degrade host secondary metabolic toxins and insecticides in insects (Berenbaum 2002, Guo et al. 1991, Sandstrom et al. 2006, Scott et al. 1998). CYP also has oxidase, reductase, isomerase, and dehydrogenase activities (Mansuy 1998).

In recent years, studies have found that CYP enzymes play an important role in the metabolism of many endogenous substances in insects. CYP has functions related to the metabolism of juvenile hormones, juvenile hormone analogues, and antijuvenile hormone substances in some insects. Moreover, many studies have shown that insect CYP is closely related to ecdysterol, fatty acid metabolism, and the synthesis of hydrocarbons in insects. Because CYP has functions in detoxification in most organisms, the CYP enzyme system is often classified as a detoxifying enzyme system.

CYP was first discovered in mammalian liver microsomes. This enzyme has been found in many eukaryotes (such as animals, plants, and fungi) and many prokaryotes (such as bacteria). CYP not only exists in different organisms, but various CYPs have also been identified in different tissues of the same species (Omura 1999). For example, CYP is abundant in the insect midgut, fat body, Malpighian tubule, and other organs or tissues, even including the head. Studies have shown that the sixth instar larvae of H. armigera have the highest CYP content in the midgut, followed by the body fat and the body wall (Qiu and Leng 1999). Insects rapidly metabolize ingested toxic substances (including plant secondary metabolites) through a detoxifying enzyme system concentrated in the midgut. The fat body is beneficial for the metabolism of toxic substances that enter the body through the epidermis or trachea (Yu et al. 2002). CYP LPR is encoded by CYP6A1 and CYP6D1 in houseflies and could be induced by phenobarbital and piperonyl butoxide but not by naphthalene, cyclopentadienes (such as dieldrin or aldrin), or beta-naphthoflavine, whereas ethanol could only induce CYP6A1 (Scott et al. 1996). Zhang et al. (2009) found that when gossypol and dimboa were added to food, the expression of CYP in Pyrausta nubilalis (Hübner) was positively correlated with the metabolism of these plant secondary metabolites (Zhang et al. 2009). Willoughby et al. (2006) used microarray technology to demonstrate that overexpression of the CYP gene plays an important role in insecticide-resistant Drosophila (Willoughby et al. 2006). Karunker et al. (2008) found that the resistance of B and Q biotypes of Bemisia tabaci (Gennadius) to imidacloprid was caused by overexpression of CYP6CM (Karunker et al. 2008). Bautista et al. (2009) found that the CYP gene CYP6BG1 was overexpressed in the fourth instar larvae of the permethrin-resistant strain of P. xylostella, and it was confirmed by gene silencing that overexpression of CYP6BG1 enhanced the metabolism of permethrin, resulting in resistance to this compound (Bautista et al. 2009). Because the role of CYP in the metabolism of endogenous compounds has been explained, the role of CYP in the metabolism of exogenous compounds has also been found. CYP drug metabolism mainly occurs in the liver and small intestine (Guengerich 2003). CYP 3A4 and CYP 2C9 had the highest expression in these tissues. Therefore, these enzymes were very common in drug metabolism (Guengerich 2003, Cao et al. 2013). It has been confirmed that CYP6AE14w was highly expressed in the midgut of H. armigera and was specifically induced by gossypol. The expression level of this gene was positively correlated with the growth of H. armigera in the presence of gossypol in food (Mao et al. 2007). CYP is involved in the metabolism of most pharmaceutical compounds (Cao et al. 2013).

In our study, 189 genes were identified in the drug metabolism-CYP pathway by transcriptome analysis, of which 29 were DEGs. Then, we found 30 DEGs for CYP from the transcriptome. The expression of these genes changed significantly under tannin stress, indicating that the CYP gene can be induced in M. troglodyta. Many studies have shown that insect resistance associated with CYP is usually characterized by overexpression of CYP, which is a common resistance mechanism (Feyereisen 1999). These results suggest that the drug metabolism pathway in M. troglodyta may mediate tannin-induced stress responses. Nevertheless, further studies are needed to determine the potential regulatory role of the drug metabolism pathway in M. troglodyta under tannin stress.